Anal. Chem. 1003, 65, 1411-1418
1411
Laser-Induced Isotopic Effects in Titanium Resonance
Ionization
R. K. Wunderlich,t G. J. WasserburgJ I. D. Hutcheon,'**and Geoffrey A. Blake8
Division of Geological and Planetary Sciences, California Institute of Technology, Mail Code 170-25,
Pasadena, California 91125
Titanium isotope ratios have been measured by
resonance ionization mass spectrometry (RIMS)
with special emphasis on the nature of laserinduced isotopic selectivity. A pronounced wavelength dependence of even mass isotope ratios is
caused by large nuclear volume effects near the
magic neutron number 28 in T i . Optical isotope
shifts, ranging from 0.07 to 0.21 cm-l, between T i
and 4eTi were measured for several transitions.
The 50Ti/4eTi and TV48Ti ratios, nevertheless,
exhibited only mass-dependent fractionation, in
which the lighter Ti isotopes were enriched by
-2.5 7'0/amu, when the laser operating parameters
were properly controlled. Odd-even mass isotopic
selectivity in the resonant ionization process was
also examined for severaltransitions as a function
of the laser polarizationstateand intensity. Under
saturating conditions for a A J = +1 transition and
a high degree of laser depolarization for a A J = 0
transition, the odd-evenisotopic enhancement was
reduced below the 2% level. The Ti isotope data
agree with our previous results for Os and indicate
that, by a careful choice of resonance transitions
and laser operating parameters, isotope ratios can
be measured accurately and reliably with RIMS.
INTRODUCTION
The high elemental selectivity1 and
of resonance ionization mass spectrometry (RIMS) gives this technique the potential to overcome many of the limitations of
secondary ion mass spectrometryand to provide a substantial
advance in elemental and isotopic in situ microanalysis. The
application of RIMS to high precision isotopic analysis is,
however, complicated by the presence of laser-induced isotopic
selectivity in the resonant photoionizationprocess. While in
some cases little or no laser-induced isotopic selectivity is
observed,5 in most studies the recorded shifts in measured
isotopic ratios, compared to the true sample composition,
+ Present address: Physikalisches Institut, Universitiit Augsburg,
Augsburg, Germany.
Also with the Charles Arms Laboratory, California Institute of
Technology.
Also with the Beckman Institute, California Institute of Technology.
(1) Hurst, G. S.; Payne, M. G.; Nayfeh, N. H.; Judish, J. P.; Wagner,
E. B. Phys. Rev. Lett. 1975,35,82-85.
(2) Parks, J. E.; Spaar, M. T.; Beekman, D. W.; Moore, L. J.; Cressman,
P. J. Resonance Ionization Spectroscopy
_ _ 1988. Znst. Phrs. Conf. Ser.
*
1988, NO.94, 197-200.
(3) Pellin,M. J.;Young,C.E.;Calaway,W.F.;Burnett,J.
W.; Jorgensen,
B.: Schweitzer. E. L.: Gruen. D. M. Nucl. Znstrum. Methods Phvs. Res.
1987, BI8, 446-451.
(4) Blum, J. D.; Pellin, M. J.; Calaway, W. F.; Young, C. E.; Gruen, D.
M.; Hutcheon, I. D.; Wasserburg, G. J. Anal. Chem. 1990, 62, 209-214.
(5) Fassett, J. D.; Powell, L. J.; Moore, L. J. Anal. Chem. 1984, 58,
'
2923-2926.
0003-2700/93/0365- 1411$04.00/0,
range from 2-3%6 to nearly 10O%,'-l3 typically with an
enhancement of the odd mass isotopes. For example, in a
recent RIMS study of Ti, laser-induced enrichments in the
47Ti/46Ti and 49Ti/48Ti ratios of up to 35% and 48%,
respectively, were observed.lOJ1 In many cases these isotopic
shifts cannot be explained solely by laser bandwidth effects
(that is, effects induced when the laser bandwidth is less than
the width of the hyperfine transition array of the odd mass
isotopes) or by selection rule differences in ionization efficiency between odd and even mass isotopes. Large shifts in
measured isotope ratios also hinder the recognition of true
isotopic abundance variations and require frequent analyses
of standards to monitor and remove laser-induced isotopic
bias."
The accuracy and reliability of isotope ratios obtained by
RIMS may thus depend criticallyon the availability of suitable
standards as well as on the relative sizes of isotopic effects
indigenous to the sample and laser-induced isotopic selectivity. Accurate Ti isotope ratios, for example, have been
determined using RIMS in the presence of large even-odd
effects by expressing the (unknown) Ti isotope ratios in the
sample relative to the same ratio measured in a standard in
the same experiment.lOJ1 However, in many important
isotope systems in geochemistry, particularly those involving
radiogenic isotopes, this ratioing technique may not be so
successful. Indeed, in the Re-Os system the ability to remove
even-odd isotope effects by referenceto a standard is limited
by the magnitude of the (poorly constrained) radiogenic
contribution to the l87Os abundance. In this case laserinduced odd-even effects must be inferred from the measured
abundance of the nonradiogenic l69Os isotope. Because of
the dynamic nature of the resonance ionization process
discussed below and the different nuclear spins of 1870s and
lMOs, correction factors obtained for lSgOsmay not be directly
applicable to 187Os.6
Three sources of laser-induced isotopic effects can be
distinguished. (i) Laser bandwidth and tuning effects: In
the presence of optical isotope shifts and hyperfine structure,
isotope biases measured with RIMS will depend on the precise
(6) Walker, R. J.; Fassett, J. D. Anal. Chem. 1986,58,2923-2927.
(7) Young,J. P.; Shaw,R. W.; Goeringer, D. E.; Smith,D. H. Resonance
Ionization Spectroscopy 1988. Zmt. Phys. Conf. Ser. 1988, No. 94,367370.
(8) Miller, C. M.; Fearey, B. L.; Palmer, B. A.; Nogar, N. S. Resonance
Ionization Spectroscopy 1988. Inst. Phys. Conf.Ser. 1988, No. 94,297306.
(9) Fairbank, W. M., Jr.; Spaar, M. T.; Parks, J. E.; Hutchinson, J. M.
Phys. Reu. A 1989,40, 2195-2198.
(10)Spiegel, D. R.; Pellin, M. J.; Calaway, W. F.; Burnett, J. W.; Coon,
S. R.; Young, C. E.; Gruen,D. M.; Clayton, R. N. LunarPlanet. Sci. 1991,
22, 1303-1304.
(11) Spiegel,D.R.;Calaway, W.F.;Davis,A. M.;Burnett,J. W.;Pellin,
M. J.;Coon,S.R.;Young,C. E.;Clayton,R. N.;Gruen,D.M.Anal. Chem.
1992,64,469-475.
(12) Feary, B. L.; Tissue, B. M.; Oliveras, J. A.; Loge, G. W.; Murrell,
M. T.; Miller, C. M. Presented at the VI International Symposium on
Resonance Ionization Spectroscopy and Ita Applications, Sante Fe, NM,
May 24-29, 1992.
(13) Hayman, C.; Comaskey, B.; Johnson, M.; Worden, W. SPZEProc.
1988,912, 200-204.
0 1993 American Chemical Society
1412
ANALYTICAL CHEMISTRY, VOL. 65, NO. 10, MAY 15, 1993
laser wavelength and resetability,12bandwidth, and intensity.
The evaluation of the wavelength dependence of isotope ratios
is a prerequisite for the measurement of reproducible and
accurate isotope ratios by RIMS and for the unambiguous
identification of other sources of laser-induced isotope effects.
(ii) Dynamic effects: Dynamic effects arise from the detailed
time evolution of the interaction between the laser radiation
and the atomic system,14-17including quantum mechanical
coherence, and lead to a dependence of measured isotope
ratios on the degree of saturation of the ionization and on the
laser pulse duration. (iii) Selection rule and polarization
effects: Selection rule and polarization effects can produce
large variations in sensitivity of the excitation-ionization
process for even vs odd mass isotopes, causing isotope ratios
to depend on the nature of the discrete transition and on the
polarization state of the laser radiation.l8J9
The processes leading to laser-induced isotopic effects have
been addressed in several recent theoreticall4-17 and experimental studies.18 Wunderlich et al.,18 for example, demonstrated several effects in the resonance ionization of Os which
can cause isotope selectivity and devised methods to reduce
them below the 1%level. Here, we apply some of these
methods to the case of Ti in order to determine whether or
not they can be generally applied to significantly reduce laserinduced isotopic selectivity in RIMS. We have chosen Ti
because of the important role Ti isotope measurements have
assumed in cosmochemistryas an indicator of nucleosynthetic
components from different stellar sources.2”-22In addition,
the large even-odd isotope shifts in a previous RIMS study
of TilOJl serve as a benchmark against which the effectiveness
of our laser tuning procedures may be evaluated.
In the following discussion, we present new data on the
resonance ionization of Ti obtained with a thermal atomization-quadrupole mass spectrometer. These data are used
to address the possible sources of laser-induced isotopic
selectivity outlined above. We then demonstrate for selected
transitions that the application of a wavelength tuning
procedure developed for Os allows measurement of eveneven isotope ratios in Ti which differ from the ratio present
in the sample only by purely mass-dependent fractionation,
that is, by fractionation which depends only on instrumental
factors other than the photoionization process.
EXPERIMENTAL SECTION
The mass spectrometer, laser system, and experimental
procedures for measuring isotope ratios used in this study were
similar to those used in our recent work on Os-RIMS and are
described in detail by Wunderlich et a1.’8 The mass spectrometer
was a Finnigan-MAT thermal ion source quadrupole mass
spectrometer modified to accommodate a laser beam in the ion
formation region. The mass resolution utilized in these experiments was typically 160(10% valley),providing a clear separation
of the Ti mass peaks. Ti samples were prepared by melting a
small amount of Ti metal powder on a Ta filament ribbon,
providing a Ti spot approximately 1mm in length at the center
of the filament. Neutral Ti atoms were emitted from the filament
at a temperatureof 1000“C, intersected by a laser beam passing
-
(14) Lambopoulos, P.; Lyras, A. A. Phys. Reo. A 1989,40,2199-2202.
(15) Lyras, A. A.; Zorman, B.; Lambropoulos, P. Phys.Reo. A 1990,42,
543-549.
(16) Whitten, W. B.; Ramsey, J. M. Appl. Spectrosc. 1990,44, 11881192.
(17) Payne, M. G.;Allman, S. L.; Parks, J. E. Spectrochem. Acta B
1991,46, 1439-1457.
(18)Wunderlich, R. K.; Hutcheon, I. D.; Wasserburg, G. J.; Blake, G.
A. Int. J. Mass Spectrom. Ion Processes 1992, 115, 123-155.
(19) Balling, L. C.; Wright, Z. Z. Appl. Phys. Lett. 1976,29,411-413.
(20) Papanastassiou, D. A. Astrophys. J. 1986, 308, L27-L30.
(21) Papanastassiou, D. A.; Brigham, C. A. Astrophys. J. 1989, 338,
L37-L40.
(22) Ireland, T. R.; Zinner, E. K.; Amari, S. Astrophys. J. 1991, 376,
L53-L56.
1-2 mm in front of the filament, and ionized (see Figure 1in ref
18for a detailed view of the ion formationregion). At this f i i e n t
temperature, the production of Ti thermal ionswas insignificant,
the thermal ion background being less than 0.1% of the Ti+
resonance ionization signal at any given Ti mass. Ion currents
were measured with an 80-MHz counting system, and isotope
ratios were collected in the peak jumping mode of the quadrupole
mass spectrometer. The linearity of the counting system was
checked by RIMS measurements of the 48Ti/46Tiisotope ratio as
a function of the intensity at mass 48. No dead-time effects were
observed for 4sTi+intensities corresponding to a maximum of
-40 photoions per laser pulse, consistent with our earlier Os
results.18
The laser system consisted of a Lambda Physik LPX llOiC
XeCl excimer laser, which pumped a Lambda Physik FL3002E
tuneable dye laser equipped with a BBO crystal for frequency
doubling. The laser system was capable of operating at repetition
rates of 200 Hz, but most experiments were performed at 20 Hz
due to increased laser instabilities at higher repetition rates. The
reproducibilityof the opticalbeam alignment in the ion extraction
region of the quadrupole mass spectrometer was assured by two
fixed apertures, which defined the laser beam position relative
to the evaporation filament. In this work we examined simple
1 + 1 (one photon resonant, two-photon ionization) schemes in
the wavelength region between 250 and 270 nm, which is the
frequency doubled tuning range of Coumarin 307 and Coumarin
153 laser dyes. Ti has an ionization potential of 6.83 eV, so that
the resonant states in this region could be ionized both by the
dye laser second harmonic and by the fundamental, which had
a nominal bandwidth of 0.2 cm-l. The dye laser could also be
operated with an intracavity etalon to reduce the bandwidth to
0.04 cm-l or, alternatively, without an intracavity etalon and in
a lower grating order to increase the visible bandwidth to 0.35
cm-l. Laser bandwidths were monitored with an etalon having
a free spectral range of 1.7 cm-1and a finesse of 18 at 500 nm.
Without the intracavity etalon no single mode structure could
be observed. The Doppler width of the transitions considered
here was <0.10 cm-l. The ability to operate the laser at increased
bandwidth was important for the control of isotope ratios in the
presence of hyperfine structure and optical isotope shifts, as will
be shown below.
The visible and UV laser beams were focused parallel to the
filament with a 50-cm focal length lens to a diameter of 0.9-1
mm. Typical pulse energies measured after transmission through
the mass spectrometer were 8 mJ in the visible and 0.4 mJ in the
UV. With a pulse duration of 20 ns, intensities of 5.1 X 107
W/cm2 in the visible and 4.5 X lo6 W/cm2 in the UV for the
ionization and resonance excitation laser fields were obtained.
These intensities saturate all bound-free transitions possessing
photoionization cross sections >2.5 X 10-l8 cm2 and discrete
where f is the transition
transitions with gf values >2 X le4,
oscillator strength and g is the degeneracy factor of the lower
state.
The presence of isobaric interferences at the Ti masses, due
to nonresonant ionization of impurities evaporating from the
filament, was checked by scanning the mass spectrometer in the
38-52 amu range with the laser tuned to the Ti 13d24s2a3F2>
13d34ps3Di> resonance at 39715.51 cm-l. No ion signal above
background was detected at masses other than those correspondingtoTiatafilamenttemperatureofllOO°C. Inparticular,
K is expected to be an abundant neutral species at this
temperature. The absence of any ion signal above noise, other
than Ti+, indicates that the nonresonant ionization of other
specieswas unimportant and that the background was dominated
by the small contribution of thermal ions. Simple 1+ 1ionization
schemes on Ti metal were chosen here to enable unambiguous
measurements and controlof laser-inducedisotopic biases. RIMS
applied to elemental and isotopic analyses of natural samples
may well require more sophisticatedionizationschemesto achieve
sufficient selectivity.
Figure 1 shows two consecutive scans of the 13d24s2a3F2>
13d34p s3Di > resonance at 39715.51 cm-l with the mass spectrometer tuned to 46 and 47 amu, respectively. The scans were
performed under nonsaturating ionization and excitation laser
power conditions and with a visible laser bandwidth of 0.2 cm-l.
-
-
ANALYTICAL CHEMISTRY, VOL. 65, NO. 10, MAY 15, 1993
t
mass 46
mass 47
3 0 0 , , , ,
I
,
I
i;ol
1419
I
4
100
-100
-200
-4
-2
0 2
A [cm-1 I
4
- 4 - 2 0 2
4
-
-
-300-'
I
I
'
I
I
'
' I
I
' ' I
I
"
A [cm-1 I
Flgurs 1. Photolonization line shape of the )3d24s2a3F2> 13d34p
s3Di> resonance transition at 39715.51 cm-l obtained with the mass
spectrometer tuned to masses 46 and 47 amu. The vlslble laser
bandwidth was 0.2 cm-1.
No hyperfine structure is revealed in the spectrum at mass 47.
Similar spectra were recorded on each transition studied in order
to ensure that all odd mass isotope HFS components could be
excited simultaneously. The line shapes are structureless and
symmetric, further indicating the absence of nonresonant ionization. Because Ti has a rather high density of states in the
250-270-nm spectral region, the dye laser wavelength was
calibrated with Re resonance ionization signals from a blank Re
filament,which provides several well-separatedtransitions. The
calibration accuracy was 0.3 cm-I.
The saturation behavior of the discretetransitionswas checked
by observing the ionization signal as a function of the detuning
of the frequency doubling crystal from the optimum phasematching tilt angle. All transitions studied could be saturated
with the UV intensities available in this experiment. The
saturation of the ionization was measuredby increasing the pump
laser intensity while tuning the frequency doubling crystal for
constant UV output at approximately 5 X lo6 W/cm2.
WAVELENGTH DEPENDENCE, OPTICAL
ISOTOPE SHIFTS, AND MASS
FRACTIONATION
Whenever the efficiency (the product of fractional ionization and instrument transmission) of mass spectrometric
analysis is less than unity, the measured ratio of ion currents
for two isotopes i and j, R$, will differ from the true
abundance ratio, R f , due to mass-dependent fractionation
arising during ionization and detection. In mass spectrometric
analysis with conventional ion sources R$ may be related to
a standard reference value, R i , by a fractionation law of the
form
The exact nature off(cu) is, in general, not known but typically
is approximated by either a linear law, 1+ mija, a power law,
(1 + a ) m l ~ ,or an exponential law, (mi/m,)a,where mij = m, m, and a is the fractionation per unit
For purely
mass-dependent fractionation, a! is mass independent and
common practice involves choosing a standard ratio R i and
calculatingthe fractionation factor a! from asuitable measured
(23) Waaaerburg, G . J.; Jacobson, J. B.; DePaolo, D. J.; McColloch, M.
T.;Wen, T . Ceochirn. Cosmochim. Acta 1981,45, 2311-2323.
(24) Hart,S. R.;Zindler, A. Znt. J. Mass Spectrorn. Zon Processes
1989,89, 287-301.
where FLand T~ are the laser photon flux and pulse duration,
respectively. If this condition can be satisfied for two isotopes
i and j in some laser tuning interval, A,and for specific values
of the laser bandwidth and intensity, then the measured
isotope ratio, R&(X), will be constant over this interval.
When the isotopes i and j have known abundancea, the isotope
ratio measured in the region over which R&(X) is constant
may be considered to differ from the standard ratio only by
purely mass-dependent fractionation. Alternatively, if the
ionization maxima of the individual (even mass) isotopes are
spectrally well resolved, the purely mass fractionated isotope
ratios may be obtained from the ratio of the ion currents
measured at the individual ionization maxima (i.e., the laser
wavelength mubt be adjusted for each isotope).
As an example, consider the laser wavelength tuning curve
for the Os A J = +1,39406.9 cm-' resonance shown in Figure
2.lS The measured l ~ O s / l ~ O
ratio
s is expressed as the per
mil (part per thousand, denoted 9bo) deviation from ita
standard valuez5using &notation. The ratio is constant over
a wavelength interval of 0.4 cm-*,indicating the simultaneous
1414
ANALYTICAL CHEMISTRY, VOL. 65, NO. 10, MAY 15, 1993
saturation of the discrete transition of both isotopes. For
larger detunings, isotope selective effects of up to 120%
become apparent with a depletion of the heavy isotope for
positive (shorter wavelength) detuning. In this experiment
the UV laser bandwidth [AoL(UV) = 0.4 cm-'1 was a factor
of 8 larger than the expected optical isotope shift.26 This
transition was used ina series of RIMS Os isotope experiments
because under saturating conditions odd-even isotopic effects
were expected to be absent. In these measurements both the
even and odd mass Os isotope ratios followed a simple mass
fractionation law in the region Rfcl(X)= constant. As pointed
out above, this law is characterized by a mass independent
a for all isotope pairs. Tuning the laser wavelength to the
center of the region with 1900s/1880s= constant enabled us
to measure the 1920s/1880s
and 16gOs/1900sratios with a
reproducibility of better than 4% (lu).18
In contrast, Figure 3 presents the wavelength dependence
of the individual ion intensities for 50Ti and 46Tiand of the
5Ti/46Tiratiofor two different AJ= +1 resonance transitions.
Throughout this paper, the Ti isotope ratios are expressed
relative to46Tias biTi = [(iTi/46Ti),,,$('Ti/46Ti)~~~
-11 X1000,
the per mil deviation from the standard values of (47Ti/46= 9.2125, ( 4 q i / 4 6 T i ) s=~ ~
Ti)STD 0.9149, (48Ti/46Ti)s~~
0.6860, and (50Ti/46Ti)s~~
= 0.6671 obtained by Niederer et
In Figure 3a, the position of the maximum signal at
mass 46 was arbitrarily identified with zero detuning, A = 0,
and the optical isotope shift between 5OTi and 46Tiis apparent
in the distinct maxima of the individual ionization line shapes.
From this measurement, we obtain AT(50-46) = 0.15 cm-',
where AT is the value of the optical isotope shift between 50Ti
and 46Ti for that transition. For a visible laser bandwidth of
AWL= 0.20 cm-l, this isotope shift produced a pronounced
wavelength dependence of the measured 5'YIW46Tiratios with
variations of more than 200% for wavelength changes as small
as 0.5Aw~near the ionization maxima. It is apparent from
Figure 3a that for this transition there is no wavelength
interval over which the 50Ti/46Tiratio remains constant, and
thus the 50Ti/46Tiratio cannot be measured accurately at a
single laser wavelength. Arbitrarily adjusting the laser
wavelength to yield h50Ti = 0 is not acceptable because this
approach would produce incorrect values for the other Ti
isotope ratios due to the offsetting effects of wavelength
dependence and mass fractionation.
The resonance transition used in the experiment shown in
Figure 3b has a smaller optical isotope shift and a larger
oscillator strength28 than the one discussed above. For this
transition there is a narrow plateau extending from A 0
-0.1 cm-l over which the 50Ti/46Tiratio remains constant,
independent of the laser wavelength. Ionization of 46Ti
occurs preferentially in this region, yielding 650Ti -1OOYi.
The width of the plateau directly shows the accuracy of the
laser wavelength setting required to obtain reproducible
isotope ratio measurements. With agfvalue of 0.4we estimate
the power broadened width of the resonance transition to lie
between 1 and 0.6 cm-l, depending on the specific MJ
component, close to that observed but large with respect to
the measured optical isotope shift of 0.07 cm-'.
In Table I we present data for the 50Ti/46Tiand 48Ti/46Ti
ratios measured in a series of experiments using different
resonance transitions and laser bandwidths. For transitions
in which the 50Ti/46Tiratio exhibited a plateau as a function
-
-
-
(25) Creaser,R. A.;Papanastassiou,D. A.;Wasserburg, G. J.Geochim.
Cosmochim. Acta 1991, 55, 397-401.
(26) Gmelin Handbook of Inorganic Chemistry, System No. 68:
Platinum; Springer Verlag: New York, 1989.
(27) Niederer, F.R.;Papanastassiou,D.A.; Wasserburg, G.J. Geochim.
Cosmochim. Acta 1981, 45, 1017-1031.
(28) Experimental Transition Probabilities for Spectral Lines of 70
Elements. NBS Mongr. (US.)
1962, No. 53.
500
t
*0°
I
,
,
,
-0.48
-0.32
-0.16
,
,
OD
0.16
Deluning (cm-1)
,
,
0.32
0.48
{
0
1("/04)
-300
P T ,
-0.18 0.00 0.18
0.36 0.54
Detuninq(cm-lJ
Flgure 3. Wavelength dependence of the ion signal for 46Tiand 5oTi,
curves A and B respectively, shown on the left-hand ordinate and of
the 5oTi/46Tiratio, curve C, on the right-hand ordinate. Typical error
of the individualdata points is 1% < 1u < 1.5% . (a)The 13d24s23F2>
13d34p s3Di > , 39715.5 cm-', resonance transitlon; oscillator
strength 0.04, AwL(UV) = 0.40 crn-'. (b) The (3d24s23F2> 13d24s4p u3Di > ,38159.5 cm-I resonance transition; oscillator strength
0.5, AwL(UV) = 0.70 cm-'.
-0.54
-0.36
-
-
of laser wavelength, isotope ratios were calculated from ion
current measurements a t a laser wavelengthfixed in the region
R$X) = constant. These data are indicated as "R(X) =
const" in Table I. For transitions where the condition R&
(A) = constant could not be met, e.g., transitions shown in
Figure 3a, isotope ratios were calculated from the ion
intensities measured at the ionization maxima of the individual isotopes. These data are indicated as "peak/peak" in
Table I. A comparison of the data in Table I shows that both
approaches yield equivalent 50Ti/46Tiratios, justifying the
assumption that the value of the 50TV4'jTiratio in the plateau
region (i.e., 650Ti -100%) differs from the true value only
by mass-dependent fractionation.
The mass fractionation per mass unit, a,was obtained from
the measured 50Ti/46Ti and 48Ti/46Tiratios, assuming f ( a )
was represented by a power law. Within the measurement
error, the CY values calculated from the 50Ti/46Ti and 48Ti/46Ti
ratios, shown in columns 3 and 4 of Table I, are in good
agreement and define a mean value, a = 0.025 f 0.002 amu-l.
In the remainder of the discussion, we use the value of CY =
0.025 amu-l to correct all measured Ti isotope ratios for mass-
-
ANALYTICAL CHEMISTRY, VOL. 65, NO. 10, MAY 15, 1993
Table I. 5oTi/4sTiand aSTi/‘sTi Ratios Measured with Different Resonance Transitions
soTi/4Ti (*la)
resonance state (cm-l)
a x 102 (flup
* T i / 4 T i (*la)
37654.8
37654.8
37654.Sh
38159.5‘
39715.5
39715.5
39686.1
0.599 (0.005)
0.604 (0.006)
0.610 (0.007)
0.603 (0.007)
0.593 (0.007)
0.607(0.009)
0.613(0.010)
0.605 (0.007)
avf
a
8.791 (0.14)
8.782 (0.14)
8.727 (0.20)
8.886 (0.16)
8.552 (0.20)
2.7 (0.2)
2.5 (0.2)
2.3 (0.3)
2.6 (0.3)
2.9 (0.3)
2.4 (0.4)
2.2 (0.4)
2.5 (0.2)
k?
8.766 (0.22)
8.751 (0.11)
a x 102 (*lay
2.4 (0.7)
2.4 (0.7)
2.7 (0.8)
1.8 (0.8)
3.8 (0.9)
g
2.5 (1.0)
2.6 (0.7)
1416
modec
R(h) = constd
R(h) = constd
peak/ peake
R(h) = constd
peak/peakd
peak/peakd
peak/peake
a calculated from measured 50TV46Tiratios assuming a power law mass fractionation correction, i.e., a = [(50Ti/46Ti)s~~/(50Ti/4Ti)meas]0.25
- 1; (5@l?i/46Ti)s~~
= 0.6671.27 a calculated from measured 48Ti/46Tiratios assuming a power law mass fractionation correction, a = [(48Ti/
46Ti)s~~/(48Ti/46Ti)mess10.5
- 1;( 4 8 T i / 4 6 T i ) =
s ~9.212~5.~~
~
R(h) = const indicates isotope ratios measuredat a fixed wavelength for each transition.
Peak/peak indicates wavelength tuned to ionization maxima for each isotope. AwL(visib1e) = 0.35 cm-l. e AhwL(visib1e) = 0.20cm-l. f Average
values calculated from the results of individual experiments. g Not measured for this transition. Transition shown in Figure 3b. Transition
shown in Figure 3a.
dependent fractionation (arising from, e.g., instrument transmission, Rayleigh distillation, etc.). The ability to measure
CY reproducibly in different experiments, including changes
of laser dyes and optical realignment, is due to the well-defined
spatial position of the ionization volume in the ion extraction
region of the ion source, as described earlier. These results
indicate that resonance transitions and laser operating
conditions may be found for which laser-induced isotopic
selectivity is absent for the even mass Ti isotopes. Further,
in the wavelength regions over which the 60TV46Tiratio is
independent of wavelength, the fractionation law must also
be valid for the odd mass Ti isotopes. Any deviations in
even-odd Ti isotope ratios, measured in these wavelength
regions, remaining after correction for instrumental maw
fractionationmust, therefore be due to intrinsic, laser-induced
even-odd effects in the resonance ionization process.
As Figures 2 and 3 demonstrate, the wavelength effects in
Ti are significantly different in sign and magnitude from the
effects observed in Os. The absorption maxima of the heavier
Os isotopes are shifted to longer wavelengths. This “normal”
shiftreflects the increaseof the average squared nuclear charge
radius in Os upon addition of neutrons.29 In Ti, the
wavelength shift is reversed. The enrichmentof “z‘imeasured
at shorter wavelengths (Figure 3) reflects a decrease in the
nuclear charge radius approachinga closed neutron shell with
magic neutron number 28.30
The field shift contribution to the IS is proportional to the
change of the electron charge density at the nucleus in the
relevant electronic states. Thus, optical isotope shifts are
most pronounced in transitions involving s electrons. Ti has
the ground-stateelectron configuration 3d24s2, the lowest term
is 13F>. The resonant excited state in the experiment shown
in Figure 3a has the electron configuration and level 13d34p
s3Di > , while the resonant state used for the experiment
shown in Figure 3b has the electron configuration and level
13d24s4p u3D; > Optical isotope shifts were much more
pronounced in the first case because the transition involves
the removal of two s electrons. Table I1 shows a list of the
optical isotope shifts AT(50-46) measured for several resonance transitions. Note the difference in the IS between the
3d34p and the 3d24s4p excited-state configurations.
The wavelength dependence of Os isotope ratios is less
pronounced compared to that observed in Ti because (a) the
change in the average nuclear charge radius with neutron
number is larger for Ti29,30
and (b) the resonance transitions
we studied in Os involve 6s electrons, which possess a much
lower electron density at the nucleus compared to those
studied in Ti. We expect these effects to be less significant
.
Table 11. Optical Isotope Shifts between 5oTi and 4sTifor
Different Resonance Transitions
resonance
frequency (cm-l)
37654.8
378f12.5~
38159.7
39686.1
39715.5
excited-state
configuration and level
13d34pu3F0>
13d34p u3F,>
8
13d24s4p u3D,0>
13d34p s3D0>
13d34p
isotope shift (cm-l)
AT(50-46) (flu)
0.21 (0.03)
0.14 (0.02)
0.07 (0.02)
0.18 (0.02)
0.15 (0.02)
a Transition from the la3F3), 170cm-l excitation, ground-state spinorbit level. All other lines involve the la3Fz) true ground-state spinorbit level.
in the Ti RIMS measurements by Spiegel et al.lOJ1 due to the
much larger Doppler width induced by secondary neutral
atom sputtering in their experiments and their selection of
an excited 3d24s4pelectron configurationin the f i t resonance
step.
ODD-EVEN ISOTOPE EFFECTS
Several physical effects can produce different ionization
efficiencies of atoms with and without nuclear spin in a
resonance ionization process. In this discussion, we assume
AWL > AWHFS, where AWHFS is the spectral width of the
hyperfine array. We distinguish dynamic effects, including
quantum mechanical coherence,lPl’ from selection rule and
polarization effecta.gJoJ8Jg Some of these effects can be
visualized by the linkage diagramsshown in Figure4a,b which
represent J = 0 * J’= 1transitions for an atom with nuclear
spin I = 0 and I = 1/2, respectively. The angular momentum
states of the atom with nuclear spin are represented as usual
by IFMF>,with F = J + I. Consider a resonance ionization
process with a pulse duration T ~ an
, inverse ionization rate
~ / R I > AWHFS,
>
and a laser bandwidth AWL > AWHFS. Under
these conditions, which are met in many resonance ionization
experiments, equilibration of ground- and excited-state
populations occurs early in the laser pulse for all allowed
transitions. From the A J = +1linkage diagram in Figure 4a,
it is easy to see that, for linearly polarized light and I = 0, half
of the total atomic population will be in the excited state,
from which ionization proceeds at a much slower rate.31 Due
to the increase of available angular momentum states, the
correspondingexcited-state population for I = 1/2increases
to two-thirds. If the ionization step is not saturated, i.e., 7 8 1
< 1, the number of ions generated from atoms with and
without nuclear spin can be considerably different with an
enhancement of the odd mass isotopes. Under saturating
ionization conditions, this difference tends to zer0.14J7J8 In
v
(29)Wohlfahrt,H.D.;Shera,E.B.;Hoehn,M.V.;Yamazaki,Y.;Steffen,
R. M.Phys. Rev. C 1981,23,533-548.
(30)Hoehn,M. V.;Shera,E. B. Phys. Rev. C 1979,20,1934-1941.
(31)Hurst, G. S.; Payne, M. G.; Kramer, S. D.; Young, J. P. Rev. Mod.
Phys. 1979,51,767-819.
1418
%
ANALYTICAL CHEMISTRY. VOL. 65. NO. 10. MAY 15, 1993
4
0
.I
-In
I
-
*m
-lb
*3n
4. (a)Linkage dlagam for a J = 0
J' = 1 dipole rmnsnlon
for atoms wlth (I = 112) and wimout (I = 0) nuclear spin. n , l n ,
repfesemathsfraclional, steady-sbte satuatedaxcitedstatepopulatkn
for excltatfon wlth linearly poladzed light. n,is the number density
of a specificisotope before onset of Interaction. m e laser bandwidth
islargerthanthesplffling betweentheF= 112andF=3/2HFSstates.
(b) Linkage dlagram for a J = 1 J' = 0 dipole transition for atoms
wlth (I= t/2) and wnhoul (I= 0) nuclear spin. n,l&,, represents the
fractional saturated ionizatlon yield for excltation with linearly polarized
light.
is the number density of a specific Isotope before onset of
Interaction.
-
Table 111. "TV(Ti and '%/Vi Ratios for Three Different
A J = +I Transitions
resonance
Ptate. d a d )
3Fr 38544.4
"4,37852.5
3D3, 38159.5
WTi I*l&
-14 (12)
35 (11)
60 (15)
Pl'i (+lnl*
3 (11)
5 (11)
144 (23)
modab
saturation
saturation
nonsaturation
L1 6'Ti (b)
= I ( ' T i / ' G T i ) , ~ ( ' T i / ~ i )-s ~
11 X 1ooO. The (''Ti/
'Wsmand (4%/qi)sn, ratiw are 0.9149 and0.6860,respctively."
Indicates if the ionization step was saturated, Le., 7 8 1 > 1.
addition, the photoionization cross sections of the individual
!JI F MF>and Ir. S J MJ> components of the excited stam
will be somewhat different and require saturation of the
ionization process in order to avoid laser-induced odd-even
isotopic bias.
For the measurement of odd-even isotope effects, the laser
wavelength was tuned such that the 45Ti146Ti ratio had the
purely mass fractionated value given in Table I in order to
assure that this ratio was free of laser wavelength tuning
effects. Table 111 shows the results of measurements of the
'-Ti *Ti and 'gTi/'Ti ratios (expressed as 64'Ti and 6 4 T i ,
respectively2'). measured on three W = + I transitions. The
ratioshave beencorrected for instrumentalmass fractionation
as described above, and we verified that the hyperfine
structureoftheodd massisotopescouldnotberesolvedwithin
the laser bandwidth in all experiments. With the exception
of the 64'Ti value of 35% for the second transition listed, the
data collected under saturating ionization conditions show
little evidence of sizable odd-even effecta. In contrast, the
data obtained under nonsaturating conditions (last row of
Table I111 exhibit large deviations from normal values for
both the "Ti q i and T i V i ratios, 6"Ti = 60 15% and
6"Ti = 144 f 23%. These laser-induced effects far exceed
theanalyticalerrorandaresimilartoisotopeshiftsfoundfor
other systems under nonsaturating conditions.3
*
We note that the difference between 647Tiand 6 4 T i shown
in the last row of Table III may he due to the larger number
of excited-state components for '
T
i (I = 5/2), i.e., 42,
compared to 28 for 47Ti(I= 3/2). In a steady state,more ' T i
than "Ti atoms are in the excited state. This is analogous
to the case shown schematically in Figure 4a. The Ti data
demonstrate that odd-even laser-induced isotopic selectivity
can be reduced below the 2 % level in the case of a 1 + 1
ionization scheme for selected AJ = + 1 transitions and
linearly polarized light. The same principles should apply to
threephoton, two-color resonant ionization schemes,although
the details of the dynamical equations are much more
complex.l5
Payne et
recently predicted that, for realistic spatial
intensity distributions of the ionizing laser radiation, some
odd mass isotope enrichment due to photoion production in
regions where theionization isnotsaturatedwillalwaplwaysoccur.
This enrichment was calculated to be approximately 3% for
Sn, depending on the detailed laser beam parameters. It
may be possible to avoid this residual odd-even effect by
using specially designed ion extraction optics, laser detuning,17,32or fast photoi~nization.~~
It also appears that,
whenever the radius of the ionization laser beam is larger
than that of the excitation laser, the ionization can be
saturated in all regions where atoms are in an excited state.
In this case, precise beam tailoring may provide another
technique to avoid isotopic selectivity related to edge effects.
In the case of fast ionization, Le. Rr >> A w m and/or if T~
< AWHFS-',the ionization process may be visualized as
originatingfrom a coherent superposition of the excited-state
hyperfine components.33-3s This situation is closely related
to quantum beat spectroscopy. For a time smaller than the
hyperfine couplingperiod, the whole excited-state population
resides in one specific Fstate with half the atomic population
in the excited state from which it will be ionized. Thus, with
regard to population balance, excitation and ionization
proceedasifthere wereno hyperfinestructure.'4J'."
Isotopic
selectivitymay, therefore, be avoided by choosing interaction
times shorter than the hyperfine coupling time, ~/AWHFS.
However, in many practical applications it is difficult to
saturate the ionization in time intervals shorter than 10-9 s.
Theresultinghghpeakpowerandincreasedlaser handwidth,
even for transform-limited ultrafast pulses, may also reduce
significantly the selective ionization properties of RIMS in
complex natural samples where a variety of elements are
present.
Frequency domain coherent techniques may provide another route to the elimination of isotopic selectivity in RIMS,
but require very complex lasers. Alternatively, following the
linkage diagram presented in Figure 4a, we suggest that
depolarization of the laser radiation, producing equal excitedstate populations for the even and odd mass isotopes, could
offer a simple means of reducing even-odd effects for A J =
+1 transitions. Selection rule and polarization effects for A J
= -1 transitions may be visualized using the linkage diagram
ahown in Figure 4b, which presents the specific J = 1 J'
= 0 case. For linearly polarized light, 67% of the groundstate population of the even mass isotope, I = 0, does not
interact with the laser field, while for the odd maas isotope
-
(32)Brandon, W.; Alman, S. L.; Payne,M. G.; Garrett, W.R.; Chen,
C. H. Presented at the VI International Symposium on Resonance
Ionization Spectroscopyand its Applications,Sank Fe. NM, May 24-29,
1992.
(33)Georges, A. T.; Lamhropoulas, P. Phys. Rev. A 1978,18,10721078.
(34)Lemhs,G.;Smith,S.J.;Khawaja,E.;Walther, H.Opt. Commun.
1979,31,313-316.
(35)Dueae,T. W.: Littman, M. G.;Zimmerman,M. L.Phys. Reu.Lett.
1975,35,1742-1754.
(36)Shore, B. W.; Johnson, M. A. Phys. Re". A 1981,W,1608-1610.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 10, MAY 15, 1993
with I = 1/2, the corresponding number is 33 % , leading to
a 100%enrichment of the odd mass isotope during ionization.
In Figure 4a,b the different HFS componentsof the resonance
transition are indicated by individual lines, but we emphasize
that the laser spectrum overlaps these transitions continuously. If the laser field has a multimode structure, several
HFS ground-statecomponentscan, in principle, interact with
a common excited state via different laser modes. This
situation has recently been studied theoretically and experimentally by Whitten and Ramsey16 using CW excitation of
Na atoms. Coherent population trapping reduces the excitation of the isotope with nuclear spin when a certain velocity
group is simultaneously resonant with two or more HFS
ground-state components. The size of the coherent dip in
the resonance excitation spectrum depends critically on the
laser bandwidth, and this effect will be absent with large
bandwidth pulsed laser excitation.
Excitation with depolarized light can avoid this type of
odd mass isotope enrichment because in all cases one-fourth
of the atoms of each isotope will be in the excited state from
which they are ionized. In many practical situations the
polarization state of the laser is not well characterized, and
the measured isotope ratios will depend on the detailed spatial
intensity distribution of the different polarization components
and the saturation behavior of the transitions associated with
these components.'S
In order to address these concerns, we have examined the
647Ti and 649Ti values for the A J = 0,13d24s2a3F2> 13d34p
u3F; > 37654.8 cm-l transition with two different degrees of
depolarization of the laser beam. The degree of depolarization, Pell, is defined as the ratio of the maximum to the
minimum intensity in the laser radiation, measured as a
function of the rotation angle of a linear polarizer. The
depolarization was generated by passing the laser beam
through a stressed quartz window, thus inducing a random
amount of birefringence across the laser beam waist. The
polarization measurements were performed for saturating
laser intensities with regard to excitation and ionization, but
the Stokes parameters of the laser fields could not be
measured. A laser wavelength setting which reproduced the
purely mass fractionated 48Ti/&Tiratio was used. These laser
tuning conditions yielded the maximum signal for 47Ti,
assuring that the 47TV46Ti ratio is free of laser wavelength
tuning bias.
For a high degree of linear polarization, the data show a
large odd mass isotope enrichment, which is greatly reduced
when the laser radiation becomes more depolarized. Specifically, with Pel,= 220 and AwL(UV)= 0.7 cm-l, we found 647Ti
= 137 f 10% and a49Ti = 135 f 10%. Two separate
experiments with Pell = 23 yielded 647Ti = 2.7 f 10% and 16
f 15%. These latter measurements were carried out with a
smaller AwL(UV)= 0.4 cm-l and with the laser wavelength
tuned to yield the maximum signal at mass 47. The fi49Ti
values, therefore, show additional wavelength-induced isotopic bias and cannot be directly compared with 647Ti. We
wish to point out, however, that the small shift (A = 0.09
cm-1) in laser wavelength between the two experiments is
considerably less than the observed IS of 0.21 cm-1.
While a beam with Pell = 23 is clearly different than a truly
depolarized beam, if saturation is achieved in both polarization
states, the results are equivalent to the ideal case of Pell = 1.
Indeed, the reduction of the odd mass isotope enrichment of
Pee = 23 can be qualitatively explained by the saturation of
the MJ = 0 MJ. = f l transition due to the presence of
approximately5% circularly polarized light in the laser field,
where J and J' represent the total angular momentum of the
ground and excited state. Consider a Rabi frequency %connecting the MJ = 1 and MJ, = 1 states with linearly polarized
-
-
1417
light. We obtain S ~ L= 1.01 X lo7 (JL)'/~
rad/s, where J L is
the linearly polarized laser intensity (in W/cm2). For the
corresponding Rabi frequency nc connectingthe MJ = 0 and
MJ. = fl states with circularly polarized light, we calculate
= 1.43 X lo7 (Jc)1/2rad/s. An oscillator strength of 5.8 X
has been used.28 The transition rate can be expressed
as R = Q2/27rI'~,where R and the laser bandwidth r L have
units of s-1 and rad/s, respectively.31 With AwL(UV)= 0.4
cm-l, we find RL = 1.35 X lo3J L s-1 and Rc = 2.71 X 103 JC
s-l. The discrete transitions are saturated if RL,CQ> 1. Under
our typical conditions of a UV intensity of 2.5 X lo6 W/cm2,
Pen = 23, and RCTP= 6, the transition driven by the minor
circularly polarized UV component is saturated. This is no
longer the case when Pel]is as large as 220.
However, even for Pell = 220 there will be some excitation
of the AMJ = f l transitions, reducing the odd mass isotope
enrichment compared to that predicted for purely linearly
polarized light. The size of the effect depends on the laser
intensity and the oscillator strength of the transition. For a
truly quantitative evaluation, the spatial intensity distribution
of the laser beam must betaken intoaccount.18 More complete
studies, in which the polarization state of the laser may be
continuously and precisely varied, will be necessary to
characterize fully the potential of depolarized light sources
in RIMS, but our results suggest that they may offer a
convenient means of reducing laser-induced isotopic bias.
In a AJ = 0 transition with integer electron angular
momentum and with linearly polarized light, odd-even
isotopic selectivity is caused by the forbidden MJ = 0 M J ~
= 0 transition.'3 This selection rule is not applicable for the
half-integer totalangular momentum of the odd mass isotopes.
The expected odd mass isotope enhancement for this transition, calculated from the corresponding linkage diagram
and PBll = m, is 647Ti = 250%. This value should be
independent of nuclear spin (i.e., 649Ti should equal 647Ti).
The observed isotope shift is smaller than expected, due to
the partial saturation of the AMJ = 0 transition by a small
degree of depolarization(or elliptical polarization) in the laser
radiation. The reduction of the odd-even isotopic selectivity
for a &J = 0 transition with increasingdegree of polarization
is consistent with our Os RIMS isotope datal8
The absence of laser-induced isotope selectivity for the
47Ti/46Tiand 49Ti/46Ti ratios under saturating ionization
conditions found here is distinctly different from the large
enhancements of 47Tiand 49Tireported by Spiegel et al.,'OJ1
who utilized a three-photon, two-color resonance ionization
scheme with an autoionizing state and a series of A J = 0
discrete transitions. While the measured 46Ti/48Tiand
5OTi/48Ti ratios are close to the expected values and exhibit
no consistent mass-dependent fractionation, the abundances
of 47Tiand 49Tiare strongly enhanced (e.g., 647Ti = 325% and
649Ti = 450% for a Ti metal target). These enrichments are
much larger than that expected from a series of A J = 0
transitions for linearly polarized light, i.e., 647Ti= 649Ti =
250%. Dynamic effeds, which have been treated theoretically
by Lyras et al.15 for multistep resonance ionization involving
A J = +1 transitions, should be greatly reduced under
saturating ionization conditions, as was the case in the study
of Spiegelet al. Thus, it appears possible that the autoionizing
step may itself introduce additional odd-even isotopic
selectivity. Autoionizing states are characterizedby specific
angular momentum quantum numbers, and additional isotopic selectivity could be introduced if the autoionizing step
involves a A J = -1 transition. This could also qualitatively
explain the observed difference in 647Tiand ij49Ti,with fi49Ti
> 647Ti. As autoionizingresonances are often used in RIMS
experiments for enhancement of the ionization efficien-
-
1418
ANALYTICAL CHEMISTRY, VOL. 65, NO. 10, MAY 15, 1993
cy,3,4,37,38
the possible influence of autoionizing transitions on
odd-even isotopic selectivity clearly requires further experimental investigation. Excitation of high-lyingRydberg levels
followed by pulsed-field ionization may offer a more robust
means of enhancing the ionization efficiency, while also
providing another degree of elemental selectivity.
CONCLUSIONS
We have presented an analysis of isotopic selectivity in the
1 + 1resonance ionization of Ti. The measured Ti isotope
ratios depend on four important parameters of the RIMS
process: (1)the precise laser wavelength and resetability; (2)
the type of transition with regard to angular momentum
change; (3) the degree of saturation of the transition; and (4)
the polarization state of the laser. Failure to take into account
the influence of these parameters on RIMS isotope ratio
measurements may cause large laser-induced isotopic shifts,
which may in turn hinder the application of RIMS to high
precision isotope ratio mass spectrometry. The pronounced
wavelength dependence of the Ti isotope ratios is caused by
large optical isotope shifts, reflecting the decrease in nuclear
volume approaching the magic neutron number 28 in 5oTi.
Wavelength-dependent measurements of even mass Ti isotope
ratios provide information about the isotope shifts of the
resonance transition and identify certain transitions for which
the effect is minimal. For selected transitions and with a
suitable choice of laser operating parameters, the even mass
Ti isotopes follow a simple mass-dependent fractionation
pattern; measured 48Ti/46Ti and 5TW46Ti ratios yield a
constant mass fraction factor, a = 0.025 f 0.002 amu-l. We
find that ratios of the even mass Ti isotopes can be measured
simultaneouslywhen the laser bandwidth or power broadened
width of the transition exceeds the optical isotope shift by
a factor of 8-10, Alternatively, the laser wavelength can be
tuned to the ionization maxima of the individual isotopes.12
This latter procedure does complicate the experimental
procedure, however.
With regard to laser-induced isotopic selectivity for the
odd mass isotopes, a A J = +1resonance transition and for
saturating laser intensity in the ionization step, our data
indicate (with the exception of one value for 647Ti)the absence
of significant odd-even effects.
(37)Rimke, H.; Hermann, G.; Mang, M.; Muhleck, C.; Riegel, J.;
Sattelberger, P.; Trautmann, N.; Ames, F.; Becker, A.; Kluge, H.-J.; Monz,
L.;Otten,E. W.;Rehlau,D.;Ruster, W. ResonanceIonizationSpectroscopy
1988. Inst. Phys. Conf. Ser. 1988,No. 94, 351-356.
(38)Kronert, U.;Becker, S.;Hilberath, T.; Kluge, H.-J.; Schultz, C.
Appl. Phys. A 1987,44, 339-345.
Coherent excitation with ultrafast pulses or specially
tailored nanosecond pulse sequences may yield high ionization
efficiencywith little laser-induced isotopic selectivity, but an
examination of these techniques will require considerably
more sophisticated laser technology than that utilized here.
Polarization scrambling offers another possibility for the
measurement of unbiased isotope ratios with the advantage
of simple instrumentation and a higher versatility in the choice
of a resonance ionization scheme for a particular element.
For example, with a A J = 0 transition and a depolarized laser
beam, the odd mass isotope enrichment was reduced below
the 2 76 level. This is an important step in the development
of a multielement capability in isotopic analysis using RIMS.
From our observations, we conclude that it is necessary to
study the effects described above spectroscopically for each
ionization scheme being considered for analytical purposes,
avoiding those transitions which either obviously or subtly
cause large isotopic selectivity. In particular, the effect of
autoionizing resonances on measured odd-even isotope ratios
requires careful consideration.
The results of these Ti RIMS experiments are in good
agreement with the results of our previous Os RIMS experiment. The data for both elements demonstrate that by
carefully choosing resonance transitions and laser operating
parameters, isotope ratios involving both even and odd mass
isotopes, free from laser-induced isotope selectivity a t the
level of counting statistics, can be measured using RIMS.
These two studies are among the few RIMS experiments to
demonstrate this important result, and we believe that such
techniques should enable RIMS to determine unbiased isotope
ratios in natural samples with a precision and accuracy of
better than 1%.
ACKNOWLEDGMENT
We thank D. A. Papanastassiou and P. G. Green for
assistance in these experiments and M. E. Johnson for
manuscript preparation. This work was supported by DOE
Grant DE FG03-88ER-13851 (to G.J.W.) and NASA Grant
NAGW-1944 (to G.A.B.). The laser system was obtained
through support from the David and Lucille Packard and
Alfred P. Sloan Foundations (G.A.B.). Division Contribution
5078(749).
RECEIVED
for review July 6, 1992. Accepted February 4,
1993.